NHS England
NHS.UK Reviews automoderation tool
NHS England automated patient-review moderation, then retired the tool
Problem
Patient reviews of NHS services posted on NHS.UK had to be checked by staff before publication, including for complaints, self-harm talk and personal names.
Approach
NHS England used natural language processing to check reviews before publication. It updated its data protection assessment in February 2024 for the switch from manual to automated moderation and published error rates for four detection models. It later listed the tool as retired without giving a reason or date.
Results
Takeaway
NHS England retired a tool for which it had published detection results for four models. It gave no reason, so the results do not explain the decision to stop using it.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
NHS England published false-positive and false-negative rates, recall and precision for four models used to moderate patient reviews. The self-harm detection model had recall of 0.99 and precision of 0.85, while complaint detection had recall of 0.85 and precision of 0.87. NHS England listed the tool as retired without explaining why.
Full details are in the original sources.